777 Crash and Fire at SFO

[Dispatch Dog]

Actually it could be even simpler than that. The same computer that monitors descent rate vs. distance to the ground could also monitor engine thrust and a few other parameters, and make a determination about the aircraft's energy state. Then if circumstances warrant, a two-word addition to the above message: "SINK RATE - INCREASE POWER - PULL UP".

I find it interesting that with all the spirited debate here about whether the computer(s) or pilot(s) should fly the plane, there's never much mention of what I'd consider the middle ground, which is having the automation provide advisory messages. How might British Midland flight 92 have turned out differently if some display had shown clearly and in large letters "L ENGINE SEVERE FAULT" or something to that effect?

I'd support anything that improves safety, but my point on many occasions has been the reliance on automation and warnings that can create a complacency which in turn becomes the cause of accidents.

If the indicator on BD92 was cleaerer, or even similar to the previous 737, then I agree, less probability of the outcome.

But should we really need warnings regarding the most basic of situations?
Again, my concern is that too much automation and reliance on warning systems may allow unsuitable people into the system that can only do their job when the systems work, and we know that they don't always work.

I think that one of the underlying causes of this accident is exactly that: A pilot that is only proficient when things work as expected, like ILS and auto throttles. This may have created the complacency that allowed decaying airspeed to be missed for so long.

[3WE]

But should we really need warnings regarding the most basic of situations?

Tough discussion...and throw in the risk of a false warning causing you to do something wrong, or a false warning simply distracting you.

...that being said, here's a tough one....what if the stall warning activated 5 or 10 knots sooner- not a new warning per se, but an adjustment to an old warning...

and maybe not too much of an adjusment- with a lot of planes a stall warning shortly before touchdown is expected.....

Would that have maybe cued a power up soon enough to save the day?

(Yes Gabriel, stalls are NOT a speed thing...except that there is a relationship to speed )

[Dispatch Dog]

Tough discussion...and throw in the risk of a false warning causing you to do something wrong, or a false warning simply distracting you.

...that being said, here's a tough one....what if the stall warning activated 5 or 10 knots sooner- not a new warning per se, but an adjustment to an old warning...

and maybe not too much of an adjusment- with a lot of planes a stall warning shortly before touchdown is expected.....

Would that have maybe cued a power up soon enough to save the day?

(Yes Gabriel, stalls are NOT a speed thing...except that there is a relationship to speed )

A good question but really only answered by professional pilots consensus... then scrutiny of the pilots that say yes!

My perception is heavily weighted by a longing for the golden age of heroic piloting.

[Evan]

But should we really need warnings regarding the most basic of situations?

Yes, because if they ever heard it, they would need to hear it. It would be a warning no licensed pilot should ever hear.

sort of like: TOO LOW GEAR...

I find it interesting that with all the spirited debate here about whether the computer(s) or pilot(s) should fly the plane, there's never much mention of what I'd consider the middle ground, which is having the automation provide advisory messages.

We have had a lot of spirited debate on the PARTNERSHIP between man and machine, which, in today's cockpit, is what it has to be. There is a lot of work being done on this new phenomena: the workplace interaction of human and artificial intelligence.

[TeeVee]

the workplace interaction of human and artificial intelligence.

which is, unfortunately, being made into a game of how little certain human beings can work and rely on the automation instead.

[3WE]

which is, unfortunately, being made into a game of how little certain human beings on rare rare rare rare rare rare rare occasions can work and rely on the automation instead.

Fixed.

[Highkeas]

Apparently there is an autopilot mode called FLCH (flight level change) on the 777; if this is used then the engines remain at idle despite the pilots having set the autothrottles to maintain a certain speed.

[EconomyClass]

So just to clear up my mind as is necessary from time to time: Pilots don't watch altitude as they approach a landing? Pilots don't know at what altitude they should be at, say, they reach the beginning of the runway?

[snydersnapshots]

Brian, Just to set your mind at ease (A little anyway). Most of us (Older) pilots hand fly plenty. If I am doing a departure out of let us use for an example HSV (Huntsville Alabama), there is no SID for this airport. Departure is runway heading to 5000’ expect radar vectors. I routinely depart HSV in the 747-8 at very close to max gross T.O. weight (987000 lbs) on the way to LUX. I will brief the departure as hand flown and request “Heading select” at 400’. Sometimes up to 12 to 14000’, sometimes into the mid 20’s. There are many airports that this will work for that we fly out of every day, ANC, MIA just to name a few.

Now on the other side of the coin, LHR, STN, FRA, AMS, these are not only some long and often complicated SID’s, but are also noise sensitive airports. Every noise violation at any of the above mentioned airports result in a $ 10000.00 U.S.D. fine! This is the time to let the automation do its magic and engage the autopilot at 250’.

The same holds true for arrivals, if it is a long complicated STAR, let the machine do its thing. Kick off the autopilot on a 5 mile final and do the landing. Weather not the best or pure dog s**t, let the airplane land itself. Weather is great (Like it was in SFO for these guys) not to tired, kick it off and hand fly the approach and the landing. I have been known to, and seen many of my colleagues hand fly the airplane from altitude all the way to the landing. Got to have fun when you get the chance!

BB...I agree completely with your philosophy. Believe it or not, several years ago I was told by a 777 guy that "you can't hand fly an airplane this big--it's too rough on the passengers." Maybe for his passengers, but if you do it right and are smooth enough, they don't even know.

A couple of months ago I had a leg from Houston to San Antonio and I briefed the F/O that I was going to hand fly the entire leg. His reaction was reminiscent of the scene toward the end of Top Gun where Maverick tells Merlin "I'm gonna bring that MiG in close:" "You're gonna do WHAT?!" I hand flew to altitude (FL220, so below RVSM airspace), turned the autopilot on to set up and brief the approach, and then hand flew the arrival and approach. The most fun I've had in an airplane since I left the Metroliner!

[Evan]

Apparently there is an autopilot mode called FLCH (flight level change) on the 777; if this is used then the engines remain at idle despite the pilots having set the autothrottles to maintain a certain speed.

"Flitch"

AFAIK, on the 777, FLCH is a pitch mode (FLCH SPD). It will quickly bring the A/T to IDLE and then put it in HOLD until the FMC target speed is reached, at which point it will return A/T to SPD mode and add thrust as needed. Also, if VNAV is armed at the time, FLCH will transit to VNAV PTH when the aircraft intercepts the VNAV descent path. For this reason it is used when high and fast on the glidepath.

There are two stealth factors with this. Firstly, if you disconnect A/P before the target speed is reached (or before reaching the VNAV descent path), the A/T will remain in HOLD at idle beyond the target speed and until the stall protection kicks in. Secondly, below 100' RA the stall protection is inhibited when the FLCH mode is active.

What I think happened here is that they disconnected the A/P before the FMC target had been met (either the speed target or the decent path - I have no information on what mode was armed at the time), thus pasting the A/T in HOLD idle. They dropped below their target speed/descent path expecting the A/T to resume SPD mode at that time (which, by design it will not do in this AFDS configuration). Therefore the sink rate was not arrested in time to catch the glidepath and the speed continued to drop. The F/O (I think) noticed the speed and called for thrust but it was noticed too late to save the approach.

Some have speculated that the issue is caused by setting the MCP altitude target at 0' (to cheat the automation). Since FLCH is part of the IAS/MACH target selection on the MCP and not the ALTITUDE selection, this theory confuses me.

BB?? Does the -8 have a similar issue?

This is what they are calling the FLCH trap.

It's not a problem with the automation, it a problem with the automation operator.

[3WE]

So just to clear up my mind as is necessary from time to time: Pilots don't watch altitude as they approach a landing? Pilots don't know at what altitude they should be at, say, they reach the beginning of the runway?

Need to split some hairs- There are 'altitude call outs ' but on the other hand, for the vast majority of the approach, they could have been aiming nearly perfectly at the landing spot and there's a comment that you want your eyes OUTSIDE the cockpit, not locked on gauges during 'critical moments'....and there are differences in your "glide angle" from time to time.

That being said, it's not an all-or nothing thing- folks are supposed to quickly scan all the primary instruments and check on stuff frequently- and then get the eyeballs back outside.

In this case, I say speed is the culprit. It is harder to judge with the eyeball, and from all appearances it seems like they were basically aiming at the landing point until very late in the game...they seem to pull up, but the plane does not power up...the speed slows...

Am I making sense?...there isn't neccesarily a call out that "we're over the approach lights, we should be at 237 feet"...that can be done extremely well with eyeballs, special approach slope lights (PAPI) that were on the airport and working, and often an ILS glide slope (though not here)...lots of good ways to judge if you are on the glide path.

Speed is generally managed with automatic throttles/speed control- but you are supposed to be glancing at the airspeed frequently to know that it's on target and you don't get good speed information by looking outside the cockpit...

[TeeVee]

apparently the PAPI on their runway was out. however it was working on the parallel runway and is supposedly visible

[Evan]

apparently the PAPI on their runway was out. however it was working on the parallel runway and is supposedly visible

The PAPI and the ILS are also not aligned at SFO RWY28L (not that this should cause any problem).

[Gabriel]

apparently the PAPI on their runway was out. however it was working on the parallel runway and is supposedly visible

AFAIK, the PAPI was working on their runway (until they crashed against it, that is). The glide slope of the ILS was not working, and they knew it since they included this item in their pre-landing briefing. On the parallel runway, it was the opposite situation: The GS was working and the PAPI was not.

[Joe H]

None of the last few posts take away the responsibility of the so called 'pilots' to monitor the ASI.